Books created in collaboration with AI on My Passion are read by millions of users every month. In a column for Scroll.media, Vladyslav Shevchenko, Exclusive Content Lead at My Passion, explains how this became possible and why curiosity, rather than fear of AI, unlocks a new level of creativity for authors.

My Passion is a platform for serial online fiction created by HOLYWATER. Within its community of authors and editors, books are produced in synergy with AI, and this is exactly what sparks the most debate.

— «AI doesn’t create anything new, so your books aren’t unique.» — «This is artificiality, not creativity.» — «Anyone can write a book with GPT — it’s not real authorship.»

We hear these claims regularly. It’s easy to imagine that a single prompt — «generate a new Harry Potter where the main character is a girl» — will produce a bestseller and send J.K. Rowling to court in a panic over her copyrights. It would be convenient if it worked that way. But it doesn’t.

Even OpenAI is now hiring a content strategist — someone who understands workflows, strategy, and text production. In other words, OpenAI openly acknowledges that a machine won’t handle this alone. It requires operators with experience, taste, and a creative vision — people who can collaborate with the system and transform its output into a cohesive work.

Bots Don’t Write Novels

Authors’ tools have evolved for centuries: from pen to typewriter to keyboard to language models. But the idea, the vision, and the integrity of the characters have always belonged to people. AI doesn’t replace the idea — it simply accelerates the path toward it. It helps find references, detect logical gaps, and check pacing or style. Remove the author, and you are left with mechanical text that lacks depth and repetition. Human creativity remains central, not optional.

There is a large team behind every My Passion book, comprising content strategists, authors, and release editors. We borrowed the best practices from classic publishing, combined them with technology (data, testing, iteration), and added AI as a co-creation tool.

«Just write a book with AI» is almost always bad advice. Quality emerges only when the process is structured, and the model’s involvement has a clear purpose: to clarify, suggest, or expand.

How We Create Bestsellers

Stage 1. Idea and strategic brief. We define the core idea, world, conflicts, and «book formula.» This stage is fully human-driven. Here, AI functions like a «living library,» helping search for references, characters, and conflicts.

Stage 2. Story backbone. Once the plot is structured, authors use trained models as «editors» to spot illogical character actions, weak plot points, and pacing issues.

Stage 3. Writing in synergy. The most AI-powered stage. With the full picture defined, AI helps generate high-quality text with a focus on emotions or sensory detail.

Stage 4. Editing. Final proofreading is fully performed by humans because this work requires emotional nuance, tone, and overall perception. If needed, the editor can ask the model to suggest alternative phrasing or flag stylistic issues, but humans remain in control of the final output.

Where AI Truly Helps

Here are the roles where AI performs best:

• «Hands.» Quickly drafting scenes based on the author’s vision.
• «Memory.» Finding effective tropes, structures, and patterns used in global bestsellers.
• «Mentor.» Highlighting weaknesses from a given perspective («as if seen through the eyes of Gabriel García Márquez or Rebecca Yarros»).
• «The most conscientious student.» Iterating over weak points: character arcs, dialogue, motivation, conflict.

Where AI Fails

Ideas generated by AI only become strong when the model has enough context. Models don’t create unique concepts from scratch — they retrieve or refine information.

We tested the «maximum freedom for AI» approach, allowing it to develop plots and character voices on its own. It resulted in repetition, plot holes, and even mixed-up character names. AI cannot replace the author. Like any tool, it has limitations.

How We Evaluate Whether a Story Is Strong Enough for Release

Before publication, each story passes through three layers of verification:

• Uniqueness of the text
• Editing
• Content strategist’s assessment — market relevance, positioning, promo strategy

We evaluate tropes, genre prospects, and marketing potential already at the brief stage.

We actively search for fresh, unexpected plots and test many stories. Sometimes a «classic» motif — say, a modern Cinderella — appears outdated at first but becomes a hit within a month. When choosing between «test more» and «perfect one book,» we always choose to test more.

We evaluate a story’s potential in the first week and throughout the month — using monetization, retention, conversion metrics, and activity in comments and communities. We even track indirect signals, like pirate requests.

Recommendations: How AI Boosts Creativity

1. Train AI for your personal needs. Create stylistic templates the model will use for descriptions, dialogue, or entire scenes.
2. Use AI as a creative mentor — especially during writer’s block or when evaluating why a text doesn’t work.
3. Eliminate fear of the blank page: start with an AI-generated draft and refine it.

Conclusion

The myth that «AI-written books = low quality» persists as long as AI is treated as a «magic button.» We don’t have a button. We have a team, a process, and a culture. AI accelerates, highlights, and suggests, but the main creator is the human. That’s why stories on My Passion resonate deeply and form fan communities.

Over the past year, we’ve independently published more than 300 online books — timely exclusive stories that BookTok eagerly picks up. AI-generated content becomes nonsense only when there’s no real person and no proven process behind it. Everything else is noise or untested assumptions.

Author: Vladyslava Shevchenko, Exclusive Content Lead My Passion